Reliability and Variability of Ki-67 Digital Image Analysis Methods for Clinical Diagnostics in Breast Cancer.

Ki-67 is a nuclear protein associated with proliferation, and a strong potential biomarker in breast cancer, but is not routinely measured in current clinical management owing to a lack of standardization. Digital image analysis (DIA) is a promising technology that could allow high-throughput analysis and standardization. There is a dearth of data on the clinical reliability as well as intra- and interalgorithmic variability of different DIA methods. In this study, we scored and compared a set of breast cancer cases in which manually counted Ki-67 has already been demonstrated to have prognostic value (n = 278) to 5 DIA methods, namely Aperio ePathology (Lieca Biosystems), Definiens Tissue Studio (Definiens AG), Qupath, an unsupervised immunohistochemical color histogram algorithm, and a deep-learning pipeline piNET. The piNET system achieved high agreement (interclass correlation coefficient: 0.850) and correlation (R = 0.85) with the reference score. The Qupath algorithm exhibited a high degree of reproducibility among all rater instances (interclass correlation coefficient: 0.889). Although piNET performed well against absolute manual counts, none of the tested DIA methods classified common Ki-67 cutoffs with high agreement or reached the clinically relevant Cohen's κ of at least 0.8. The highest agreement achieved was a Cohen's κ statistic of 0.73 for cutoffs 20% and 25% by the piNET system. The main contributors to interalgorithmic variation and poor cutoff characterization included heterogeneous tumor biology, varying algorithm implementation, and setting assignments. It appears that image segmentation is the primary explanation for semiautomated intra-algorithmic variation, which involves significant manual intervention to correct. Automated pipelines, such as piNET, may be crucial in developing robust and reproducible unbiased DIA approaches to accurately quantify Ki-67 for clinical diagnosis in the future.

AI improves accuracy, agreement and efficiency of pathologists for Ki67 assessments in breast cancer.

The Ki-67 proliferation index (PI) guides treatment decisions in breast cancer but suffers from poor inter-rater reproducibility. Although AI tools have been designed for Ki-67 assessment, their impact on pathologists' work remains understudied. 90 international pathologists were recruited to assess the Ki-67 PI of ten breast cancer tissue microarrays with and without AI. Accuracy, agreement, and turnaround time with and without AI were compared. Pathologists' perspectives on AI were collected. Using AI led to a significant decrease in PI error (2.1% with AI vs. 5.9% without AI, p < 0.001), better inter-rater agreement (ICC: 0.70 vs. 0.92; Krippendorff's α: 0.63 vs. 0.89; Fleiss' Kappa: 0.40 vs. 0.86), and an 11.9% overall median reduction in turnaround time. Most pathologists (84%) found the AI reliable. For Ki-67 assessments, 76% of respondents believed AI enhances accuracy, 82% said it improves consistency, and 83% trust it will improve efficiency. This study highlights AI's potential to standardize Ki-67 scoring, especially between 5 and 30% PI-a range with low PI agreement. This could pave the way for a universally accepted PI score to guide treatment decisions, emphasizing the promising role of AI integration into pathologist workflows.

Brain Age Estimation on a Dementia Cohort Using FLAIR MRI Biomarkers.

BACKGROUND AND PURPOSE: The prodromal stage of Alzheimer's disease presents an imperative intervention window. This work focuses on using brain age prediction models and biomarkers from FLAIR MR imaging to identify subjects who progress to Alzheimer's disease (converting mild cognitive impairment) or those who remain stable (stable mild cognitive impairment). MATERIALS AND METHODS: A machine learning model was trained to predict the age of normal control subjects on the basis of volume, intensity, and texture features from 3239 FLAIR MRI volumes. The brain age gap estimation (BrainAGE) was computed as the difference between the predicted and true age, and it was used as a biomarker for both cross-sectional and longitudinal analyses. Differences in biomarker means, slopes, and intercepts were investigated using ANOVA and Tukey post hoc test. Correlation analysis was performed between brain age gap estimation and established Alzheimer's disease indicators. RESULTS: The brain age prediction model showed accurate results (mean absolute error = 2.46 years) when testing on held out normal control data. The computed BrainAGE metric showed significant differences between the stable mild cognitive impairment and converting mild cognitive impairment groups in cross-sectional and longitudinal analyses, most notably showing significant differences up to 4 years before conversion to Alzheimer's disease. A significant correlation was found between BrainAGE and previously established Alzheimer's disease conversion biomarkers. CONCLUSIONS: The BrainAGE metric can allow clinicians to consider a single explainable value that summarizes all the biomarkers because it considers many dimensions of disease and can determine whether the subject has normal aging patterns or if he or she is trending into a high-risk category using a single value.

Detecting conversion from mild cognitive impairment to Alzheimer's disease using FLAIR MRI biomarkers.

Mild cognitive impairment (MCI) is the prodromal phase of Alzheimer's disease (AD) and while it presents as an imperative intervention window, it is difficult to detect which subjects convert to AD (cMCI) and which ones remain stable (sMCI). The objective of this work was to investigate fluid-attenuated inversion recovery (FLAIR) MRI biomarkers and their ability to differentiate between sMCI and cMCI subjects in cross-sectional and longitudinal data. Three types of biomarkers were investigated: volume, intensity and texture. Volume biomarkers included total brain volume, cerebrospinal fluid volume (CSF), lateral ventricular volume, white matter lesion volume, subarachnoid CSF, and grey matter (GM) and white matter (WM), all normalized to intracranial volume. The mean intensity, kurtosis, and skewness of the GM and WM made up the intensity features. Texture features quantified homogeneity and microstructural tissue changes of GM and WM regions. Composite indices were also considered, which are biomarkers that represent an aggregate sum (z-score normalization and summation) of all biomarkers. The FLAIR MRI biomarkers successfully identified high-risk subjects as significant differences (p < 0.05) were found between the means of the sMCI and cMCI groups and the rate of change over time for several individual biomarkers as well as the composite indices for both cross-sectional and longitudinal analyses. Classification accuracy and feature importance analysis showed volume biomarkers to be most predictive, however, best performance was obtained when complimenting the volume biomarkers with the intensity and texture features. Using all the biomarkers, accuracy of 86.2 % and 69.2 % was achieved for normal control-AD and sMCI-cMCI classification respectively. Survival analysis demonstrated that the majority of the biomarkers showed a noticeable impact on the AD conversion probability 4 years prior to conversion. Composite indices were the top performers for all analyses including feature importance, classification, and survival analysis. This demonstrated their ability to summarize various dimensions of disease into single-valued metrics. Significant correlation (p < 0.05) with phosphorylated-tau and amyloid-beta CSF biomarkers was found with all the FLAIR biomarkers. The proposed biomarker system is easily attained as FLAIR is routinely acquired, models are not computationally intensive and the results are explainable, thus making this pipeline easily integrated into clinical workflow.

Synthesis of diffusion-weighted MRI scalar maps from FLAIR volumes using generative adversarial networks.

Introduction: Acquisition and pre-processing pipelines for diffusion-weighted imaging (DWI) volumes are resource- and time-consuming. Generating synthetic DWI scalar maps from commonly acquired brain MRI sequences such as fluid-attenuated inversion recovery (FLAIR) could be useful for supplementing datasets. In this work we design and compare GAN-based image translation models for generating DWI scalar maps from FLAIR MRI for the first time. Methods: We evaluate a pix2pix model, two modified CycleGANs using paired and unpaired data, and a convolutional autoencoder in synthesizing DWI fractional anisotropy (FA) and mean diffusivity (MD) from whole FLAIR volumes. In total, 420 FLAIR and DWI volumes (11,957 images) from multi-center dementia and vascular disease cohorts were used for training/testing. Generated images were evaluated using two groups of metrics: (1) human perception metrics including peak signal-to-noise ratio (PSNR) and structural similarity (SSIM), (2) structural metrics including a newly proposed histogram similarity (Hist-KL) metric and mean squared error (MSE). Results: Pix2pix demonstrated the best performance both quantitatively and qualitatively with mean PSNR, SSIM, and MSE metrics of 23.41 dB, 0.8, 0.004, respectively for MD generation, and 24.05 dB, 0.78, 0.004, respectively for FA generation. The new histogram similarity metric demonstrated sensitivity to differences in fine details between generated and real images with mean pix2pix MD and FA Hist-KL metrics of 11.73 and 3.74, respectively. Detailed analysis of clinically relevant regions of white matter (WM) and gray matter (GM) in the pix2pix images also showed strong significant (p < 0.001) correlations between real and synthetic FA values in both tissue types (R = 0.714 for GM, R = 0.877 for WM). Discussion/conclusion: Our results show that pix2pix's FA and MD models had significantly better structural similarity of tissue structures and fine details than other models, including WM tracts and CSF spaces, between real and generated images. Regional analysis of synthetic volumes showed that synthetic DWI images can not only be used to supplement clinical datasets, but demonstrates potential utility in bypassing or correcting registration in data pre-processing.

Alzheimer's and vascular disease classification using regional texture biomarkers in FLAIR MRI.

Interactions between subcortical vascular disease and dementia due to Alzheimer's disease (AD) are unclear, and clinical overlap between the diseases makes diagnosis challenging. Existing studies have shown regional microstructural changes specific to each disease, and that textures in fluid-attenuated inversion recovery (FLAIR) MRI images may characterize abnormalities in tissue microstructure. This work aims to investigate regional FLAIR biomarkers that can differentiate dementia cohorts with and without subcortical vascular disease. FLAIR and diffusion MRI (dMRI) volumes were obtained in 65 mild cognitive impairment (MCI), 21 AD, 44 subcortical vascular MCI (scVMCI), 22 Mixed etiology, and 48 healthy elderly patients. FLAIR texture and intensity biomarkers were extracted from the normal appearing brain matter (NABM), WML penumbra, blood supply territory (BST), and white matter tract regions of each patient. All FLAIR biomarkers were correlated to dMRI metrics in each region and global WML load, and biomarker means between groups were compared using ANOVA. Binary classifications were performed using Random Forest classifiers to investigate the predictive nature of the regional biomarkers, and SHAP feature analysis was performed to further investigate optimal regions of interest for differentiating disease groups. The regional FLAIR biomarkers were strongly correlated to MD, while all biomarker regions but white matter tracts were strongly correlated to WML burden. Classification between Mixed disease and healthy, AD, and scVMCI patients yielded accuracies of 97%, 81%, and 72% respectively using WM tract biomarkers. Classification between scVMCI and healthy, MCI, and AD patients yielded accuracies of 89%, 84%, and 79% respectively using penumbra biomarkers. Only the classification between AD and healthy patients had optimal results using NABM biomarkers. This work presents novel regional FLAIR biomarkers that may quantify white matter degeneration related to subcortical vascular disease, and which indicate that investigating degeneration in specific regions may be more important than assessing global WML burden in vascular disease groups.

Mitosis domain generalization in histopathology images - The MIDOG challenge.

The density of mitotic figures (MF) within tumor tissue is known to be highly correlated with tumor proliferation and thus is an important marker in tumor grading. Recognition of MF by pathologists is subject to a strong inter-rater bias, limiting its prognostic value. State-of-the-art deep learning methods can support experts but have been observed to strongly deteriorate when applied in a different clinical environment. The variability caused by using different whole slide scanners has been identified as one decisive component in the underlying domain shift. The goal of the MICCAI MIDOG 2021 challenge was the creation of scanner-agnostic MF detection algorithms. The challenge used a training set of 200 cases, split across four scanning systems. As test set, an additional 100 cases split across four scanning systems, including two previously unseen scanners, were provided. In this paper, we evaluate and compare the approaches that were submitted to the challenge and identify methodological factors contributing to better performance. The winning algorithm yielded an F1 score of 0.748 (CI95: 0.704-0.781), exceeding the performance of six experts on the same task.

Impact of artificial intelligence on pathologists' decisions: an experiment.

OBJECTIVE: The accuracy of artificial intelligence (AI) in medicine and in pathology in particular has made major progress but little is known on how much these algorithms will influence pathologists' decisions in practice. The objective of this paper is to determine the reliance of pathologists on AI and to investigate whether providing information on AI impacts this reliance. MATERIALS AND METHODS: The experiment using an online survey design. Under 3 conditions, 116 pathologists and pathology students were tasked with assessing the Gleason grade for a series of 12 prostate biopsies: (1) without AI recommendations, (2) with AI recommendations, and (3) with AI recommendations accompanied by information about the algorithm itself, specifically algorithm accuracy rate and algorithm decision-making process. RESULTS: Participant responses were significantly more accurate with the AI decision aids than without (92% vs 87%, odds ratio 13.30, P < .01). Unexpectedly, the provision of information on the algorithm made no significant difference compared to AI without information. The reliance on AI correlated with general beliefs on AI's usefulness but not with particular assessments of the AI tool offered. Decisions were made faster when AI was provided. DISCUSSION: These results suggest that pathologists are willing to rely on AI regardless of accuracy or explanations. Generalization beyond the specific tasks and explanations provided will require further studies. CONCLUSION: This study suggests that the factors that influence the reliance on AI differ in practice from beliefs expressed by clinicians in surveys. Implementation of AI in prospective settings should take individual behaviors into account.

Delusional Severity Is Associated with Abnormal Texture in FLAIR MRI.

Background: This study examines the relationship between delusional severity in cognitively impaired adults with automatically computed volume and texture biomarkers from the Normal Appearing Brain Matter (NABM) in FLAIR MRI. Methods: Patients with mild cognitive impairment (MCI, n = 24) and Alzheimer's Disease (AD, n = 18) with delusions of varying severities based on Neuropsychiatric Inventory-Questionnaire (NPI-Q) (1­mild, 2­moderate, 3­severe) from the Alzheimer's Disease Neuroimaging Initiative (ADNI) were analyzed for this task. The NABM region, which is gray matter (GM) and white matter (WM) combined, was automatically segmented in FLAIR MRI volumes with intensity standardization and thresholding. Three imaging biomarkers were computed from this region, including NABM volume and two texture markers called "Integrity" and "Damage". Together, these imaging biomarkers quantify structural changes in brain volume, microstructural integrity and tissue damage. Multivariable regression was used to investigate relationships between imaging biomarkers and delusional severities (1, 2 and 3). Sex, age, education, APOE4 and baseline cerebrospinal fluid (CSF) tau were included as co-variates. Results: Biomarkers were extracted from a total of 42 participants with longitudinal time points representing 164 imaging volumes. Significant associations were found for all three NABM biomarkers between delusion level 3 and level 1. Integrity was also sensitive enough to show differences between delusion level 1 and delusion level 2. A significant specified interaction was noted with severe delusions (level 3) and CSF tau for all imaging biomarkers (p < 0.01). APOE4 homozygotes were also significantly related to the biomarkers. Conclusion: Cognitively impaired older adults with more severe delusions have greater global brain disease burden in the WM and GM combined (NABM) as measured using FLAIR MRI. Relative to patients with mild delusions, tissue degeneration in the NABM was more pronounced in subjects with higher delusional symptoms, with a significant association with CSF tau. Future studies are required to establish potential tau-associated mechanisms of increased delusional severity.

Explainable Biomarkers for Automated Glomerular and Patient-Level Disease Classification.

Pathologists use multiple microscopy modalities to assess renal biopsy specimens. Besides usual diagnostic features, some changes are too subtle to be properly defined. Computational approaches have the potential to systematically quantitate subvisual clues, provide pathogenetic insight, and link to clinical outcomes. To this end, a proof-of-principle study is presented demonstrating that explainable biomarkers through machine learning can distinguish between glomerular disorders at the light-microscopy level. The proposed system used image analysis techniques and extracted 233 explainable biomarkers related to color, morphology, and microstructural texture. Traditional machine learning was then used to classify minimal change disease (MCD), membranous nephropathy (MN), and thin basement membrane nephropathy (TBMN) diseases on a glomerular and patient-level basis. The final model combined the Gini feature importance set and linear discriminant analysis classifier. Six morphologic (nuclei-to-glomerular tuft area, nuclei-to-glomerular area, glomerular tuft thickness greater than ten, glomerular tuft thickness greater than three, total glomerular tuft thickness, and glomerular circularity) and four microstructural texture features (luminal contrast using wavelets, nuclei energy using wavelets, nuclei variance using color vector LBP, and glomerular correlation using GLCM) were, together, the best performing biomarkers. Accuracies of 77% and 87% were obtained for glomerular and patient-level classification, respectively. Computational methods, using explainable glomerular biomarkers, have diagnostic value and are compatible with our existing knowledge of disease pathogenesis. Furthermore, this algorithm can be applied to clinical datasets for novel prognostic and mechanistic biomarker discovery.

MiNuGAN: Dual Segmentation of Mitoses and Nuclei Using Conditional GANs on Multi-center Breast H&E Images.

Breast cancer is the second most commonly diagnosed type of cancer among women as of 2021. Grading of histopathological images is used to guide breast cancer treatment decisions and a critical component of this is a mitotic score, which is related to tumor aggressiveness. Manual mitosis counting is an extremely tedious manual task, but automated approaches can be used to overcome inefficiency and subjectivity. In this paper, we propose an automatic mitosis and nuclear segmentation method for a diverse set of H&E breast cancer pathology images. The method is based on a conditional generative adversarial network to segment both mitoses and nuclei at the same time. Architecture optimizations are investigated, including hyper parameters and the addition of a focal loss. The accuracy of the proposed method is investigated using images from multiple centers and scanners, including TUPAC16, ICPR14 and ICPR12 datasets. In TUPAC16, we use 618 carefully annotated images of size 256×256 scanned at 40×. TUPAC16 is used to train the model, and segmentation performance is measured on the test set for both nuclei and mitoses. Results on 200 held-out testing images from the TUPAC16 dataset were mean DSC = 0.784 and 0.721 for nuclear and mitosis, respectively. On 202 ICPR12 images, mitosis segmentation accuracy had a mean DSC = 0.782, indicating the model generalizes well to unseen datasets. For datasets that had mitosis centroid annotations, which included 200 TUPAC16, 202 ICPR12 and 524 ICPR14, a mean F1-score of 0.854 was found indicating high mitosis detection accuracy.

Machine learning in renal pathology.

Introduction: When assessing kidney biopsies, pathologists use light microscopy, immunofluorescence, and electron microscopy to describe and diagnose glomerular lesions and diseases. These methods can be laborious, costly, fraught with inter-observer variability, and can have delays in turn-around time. Thus, computational approaches can be designed as screening and/or diagnostic tools, potentially relieving pathologist time, healthcare resources, while also having the ability to identify novel biomarkers, including subvisual features. Methods: Here, we implement our recently published biomarker feature extraction (BFE) model along with 3 pre-trained deep learning models (VGG16, VGG19, and InceptionV3) to diagnose 3 glomerular diseases using PAS-stained digital pathology images alone. The BFE model extracts a panel of 233 explainable features related to underlying pathology, which are subsequently narrowed down to 10 morphological and microstructural texture features for classification with a linear discriminant analysis machine learning classifier. 45 patient renal biopsies (371 glomeruli) from minimal change disease (MCD), membranous nephropathy (MN), and thin-basement membrane nephropathy (TBMN) were split into training/validation and held out sets. For the 3 deep learningmodels, data augmentation and Grad-CAM were used for better performance and interpretability. Results: The BFE model showed glomerular validation accuracy of 67.6% and testing accuracy of 76.8%. All deep learning approaches had higher validation accuracies (most for VGG16 at 78.5%) but lower testing accuracies. The highest testing accuracy at the glomerular level was VGG16 at 71.9%, while at the patient-level was InceptionV3 at 73.3%. Discussion: The results highlight the potential of both traditional machine learning and deep learning-based approaches for kidney biopsy evaluation.

Automated Midline Estimation for Symmetry Analysis of Cerebral Hemispheres in FLAIR MRI.

To perform brain asymmetry studies in large neuroimaging archives, reliable and automatic detection of the interhemispheric fissure (IF) is needed to first extract the cerebral hemispheres. The detection of the IF is often referred to as mid-sagittal plane estimation, as this plane separates the two cerebral hemispheres. However, traditional planar estimation techniques fail when the IF presents a curvature caused by existing pathology or a natural phenomenon known as brain torque. As a result, midline estimates can be inaccurate. In this study, a fully unsupervised midline estimation technique is proposed that is comprised of three main stages: head angle correction, control point estimation and midline generation. The control points are estimated using a combination of intensity, texture, gradient, and symmetry-based features. As shown, the proposed method automatically adapts to IF curvature, is applied on a slice-to-slice basis for more accurate results and also provides accurate delineation of the midline in the septum pellucidum, which is a source of failure for traditional approaches. The method is compared to two state-of-the-art methods for midline estimation and is validated using 75 imaging volumes (~3,000 imaging slices) acquired from 38 centers of subjects with dementia and vascular disease. The proposed method yields the lowest average error across all metrics: Hausdorff distance (HD) was 0.32 ± 0.23, mean absolute difference (MAD) was 1.10 ± 0.38 mm and volume difference was 7.52 ± 5.40 and 5.35 ± 3.97 ml, for left and right hemispheres, respectively. Using the proposed method, the midline was extracted for 5,360 volumes (~275K images) from 83 centers worldwide, acquired by GE, Siemens and Philips scanners. An asymmetry index was proposed that automatically detected outlier segmentations (which were <1% of the total dataset). Using the extracted hemispheres, hemispheric asymmetry texture biomarkers of the normal-appearing brain matter (NABM) were analyzed in a dementia cohort, and significant differences in biomarker means were found across SCI and MCI and SCI and AD.

piNET-An Automated Proliferation Index Calculator Framework for Ki67 Breast Cancer Images.

In this work, a novel proliferation index (PI) calculator for Ki67 images called piNET is proposed. It is successfully tested on four datasets, from three scanners comprised of patches, tissue microarrays (TMAs) and whole slide images (WSI), representing a diverse multi-centre dataset for evaluating Ki67 quantification. Compared to state-of-the-art methods, piNET consistently performs the best over all datasets with an average PI difference of 5.603%, PI accuracy rate of 86% and correlation coefficient R = 0.927. The success of the system can be attributed to several innovations. Firstly, this tool is built based on deep learning, which can adapt to wide variability of medical images-and it was posed as a detection problem to mimic pathologists' workflow which improves accuracy and efficiency. Secondly, the system is trained purely on tumor cells, which reduces false positives from non-tumor cells without needing the usual pre-requisite tumor segmentation step for Ki67 quantification. Thirdly, the concept of learning background regions through weak supervision is introduced, by providing the system with ideal and non-ideal (artifact) patches that further reduces false positives. Lastly, a novel hotspot analysis is proposed to allow automated methods to score patches from WSI that contain "significant" activity.

Whole volume brain extraction for multi-centre, multi-disease FLAIR MRI datasets.

Automatic segmentation of the brain from magnetic resonance images (MRI) is a fundamental step in many neuroimaging processing frameworks. There are mature technologies for this task for T1- and T2-weighted MRI; however, a widely-accepted brain extraction method for Fluid-Attenuated Inversion Recovery (FLAIR) MRI has yet to be established. FLAIR MRI are becoming increasingly important for the analysis of neurodegenerative diseases and tools developed for this sequence would have clinical value. To maximize translation opportunities and for large scale research studies, algorithms for brain extraction in FLAIR MRI should generalize to multi-centre (MC) data. To this end, this work proposes a fully automated, whole volume brain extraction methodology for MC FLAIR MRI datasets. The framework is built using a novel standardization framework which reduces acquisition artifacts, standardizes the intensities of tissues and normalizes the spatial coordinates of brain tissue across MC datasets. Using the standardized datasets, an intuitive set of features based on intensity, spatial location and gradients are extracted and classified using a random forest (RF) classifier to segment the brain tissue class. A series of experiments were conducted to optimize classifier parameters, and to determine segmentation accuracy for standardized and unstandardized (original) data, as a function of scanner vendor, feature type and disease type. The models are trained, tested and validated on 156 image volumes (∼8000 image slices) from two multi-centre, multi-disease datasets, acquired with varying imaging parameters from 30 centres and three scanner vendors. The image datasets, denoted as CAIN and ADNI for vascular and dementia disease, respectively, represent a diverse collection of MC data to test the generalization capabilities of the proposed design. Results demonstrate the importance of standardization for segmentation of MC data, as models trained on standardized data yielded a drastic improvement in brain extraction accuracy compared to the original, unstandardized data (CAIN: DSC = 91% and ADNI: DSC = 86% vs. CAIN: 78% and ADNI: 65%). It was also found that models created from one scanner vendor based on unstandardized data yielded poor segmentation results in data acquired from other scanner vendors, which was improved through standardization. These results demonstrate that to create consistency in segmentations from multi-institutional datasets it is paramount that MC variability be mitigated to improve stability and to ensure generalization of machine learning algorithms for MRI.

Assessing the Impact of Color Normalization in Convolutional Neural Network-Based Nuclei Segmentation Frameworks.

Image analysis tools for cancer, such as automatic nuclei segmentation, are impacted by the inherent variation contained in pathology image data. Convolutional neural networks (CNN), demonstrate success in generalizing to variable data, illustrating great potential as a solution to the problem of data variability. In some CNN-based segmentation works for digital pathology, authors apply color normalization (CN) to reduce color variability of data as a preprocessing step prior to prediction, while others do not. Both approaches achieve reasonable performance and yet, the reasoning for utilizing this step has not been justified. It is therefore important to evaluate the necessity and impact of CN for deep learning frameworks, and its effect on downstream processes. In this paper, we evaluate the effect of popular CN methods on CNN-based nuclei segmentation frameworks.

IHC Color Histograms for Unsupervised Ki67 Proliferation Index Calculation.

Automated image analysis tools for Ki67 breast cancer digital pathology images would have significant value if integrated into diagnostic pathology workflows. Such tools would reduce the workload of pathologists, while improving efficiency, and accuracy. Developing tools that are robust and reliable to multicentre data is challenging, however, differences in staining protocols, digitization equipment, staining compounds, and slide preparation can create variabilities in image quality and color across digital pathology datasets. In this work, a novel unsupervised color separation framework based on the IHC color histogram (IHCCH) is proposed for the robust analysis of Ki67 and hematoxylin stained images in multicentre datasets. An "overstaining" threshold is implemented to adjust for background overstaining, and an automated nuclei radius estimator is designed to improve nuclei detection. Proliferation index and F1 scores were compared between the proposed method and manually labeled ground truth data for 30 TMA cores that have ground truths for Ki67+ and Ki67- nuclei. The method accurately quantified the PI over the dataset, with an average proliferation index difference of 3.25%. To ensure the method generalizes to new, diverse datasets, 50 Ki67 TMAs from the Protein Atlas were used to test the validated approach. As the ground truth for this dataset is PI ranges, the automated result was compared to the PI range. The proposed method correctly classified 74 out of 80 TMA images, resulting in a 92.5% accuracy. In addition to these validations experiments, performance was compared to two color-deconvolution based methods, and to six machine learning classifiers. In all cases, the proposed work maintained more consistent (reproducible) results, and higher PI quantification accuracy.

Measurement Variability Following MRI System Upgrade.

Major hardware/software changes to MRI platforms, either planned or unplanned, will almost invariably occur in longitudinal studies. Our objective was to assess the resulting variability on relevant imaging measurements in such context, specifically for three Siemens Healthcare Magnetom Trio upgrades to the Prismafit platform. We report data acquired on three healthy volunteers scanned before and after three different platform upgrades. We assessed differences in image signal [contrast-to-noise ratio (CNR)] on T1-weighted images (T1w) and fluid-attenuated inversion recovery images (FLAIR); brain morphometry on T1w image; and small vessel disease (white matter hyperintensities; WMH) on FLAIR image. Prismafit upgrade resulted in higher (30%) and more variable neocortical CNR and larger brain volume and thickness mainly in frontal areas. A significant relationship was observed between neocortical CNR and neocortical volume. For FLAIR images, no significant CNR difference was observed, but WMH volumes were significantly smaller (-68%) after Prismafit upgrade, when compared to results on the Magnetom Trio. Together, these results indicate that Prismafit upgrade significantly influenced image signal, brain morphometry measures and small vessel diseases measures and that these effects need to be taken into account when analyzing results from any longitudinal study undergoing similar changes.

Pathology-preserving intensity standardization framework for multi-institutional FLAIR MRI datasets.

Fluid-Attenuated Inversion Recovery (FLAIR) MRI are used by physicians to analyze white matter lesions (WML) of the brain, which are related to neurodegenerative diseases such as dementia and vascular disease. To study the causes and progression of these diseases, multi-centre (MC) studies are conducted, with images acquired and analyzed from multiple institutions. Due to differences in acquisition software and hardware, there is variability in image properties, which creates challenges for automated algorithms. This work explores this variability, known as the MC effect, by analyzing nearly 5000 MC FLAIR volumes and proposes an intensity standardization framework to normalize intensity non-standardness in FLAIR MRI, while ensuring the appearance of WML. Results show that original image characteristics varied significantly between scanner vendors and centres, and that this variability was reduced with standardization. To further highlight the utility of intensity standardization, a threshold-based brain extraction algorithm is implemented and compared with a classifier-based approach. A competitive Dice Similarity Coefficient of 81% was achieved on 183 volumes, demonstrating that optimized pre-processing can effectively reduce the variability in MC studies, allowing for simplified algorithms to be applied on large datasets robustly.

Standardized Assessment of Automatic Segmentation of White Matter Hyperintensities and Results of the WMH Segmentation Challenge.

Quantification of cerebral white matter hyperintensities (WMH) of presumed vascular origin is of key importance in many neurological research studies. Currently, measurements are often still obtained from manual segmentations on brain MR images, which is a laborious procedure. The automatic WMH segmentation methods exist, but a standardized comparison of the performance of such methods is lacking. We organized a scientific challenge, in which developers could evaluate their methods on a standardized multi-center/-scanner image dataset, giving an objective comparison: the WMH Segmentation Challenge. Sixty T1 + FLAIR images from three MR scanners were released with the manual WMH segmentations for training. A test set of 110 images from five MR scanners was used for evaluation. The segmentation methods had to be containerized and submitted to the challenge organizers. Five evaluation metrics were used to rank the methods: 1) Dice similarity coefficient; 2) modified Hausdorff distance (95th percentile); 3) absolute log-transformed volume difference; 4) sensitivity for detecting individual lesions; and 5) F1-score for individual lesions. In addition, the methods were ranked on their inter-scanner robustness; 20 participants submitted their methods for evaluation. This paper provides a detailed analysis of the results. In brief, there is a cluster of four methods that rank significantly better than the other methods, with one clear winner. The inter-scanner robustness ranking shows that not all the methods generalize to unseen scanners. The challenge remains open for future submissions and provides a public platform for method evaluation.